Coinbase Launches AI 'Teammate' CEEcil: Remembers Team Context, Proactively Follows Up, and Makes Autonomous Judgments

On August 28, Coinbase disclosed the architecture and application of its internal AI system "CEEcil." The system is not a simple chatbot but is designed as an AI agent resembling a "team member" that can be @mentioned on Slack, with long-term memory, proactive follow-up, and judgment capabilities.

CEEcil retains team context through a three-tier memory mechanism: it continuously extracts decisions, blockers, and responsible parties from authorized channels in the background; each night it consolidates short-term observations into long-term summaries; and when answering questions, it retrieves historical memory, recent conversations, and real-time data on demand. The system also employs a tiered model architecture, where simple queries are prioritized for processing via APIs or knowledge bases, with only complex, multi-step tasks delegated to stronger AI models to reduce cost and latency. Coinbase stated that CEEcil can currently help teams handle production incidents and answer expert knowledge that would otherwise require hours of waiting; in the future, it aims to expand this model into multiple AI agents tailored to specific teams, enabling AI not only to "answer questions" but also to continuously remember, proactively participate, and complete actual work.

[TechFlow]